Wuraw
Open App

DecentralizedInferenceRouter

Powered by x402

settled usdg /

  • robinhood
  • base
  • ethereum
  • arbitrum
  • optimism
  • polygon
  • avalanche
  • bsc
  • solana

+ 10 more

Robinhood

settle/usdg

USDG

Settlement [USDG/Robinhood]

Robinhood ChainNative settle
x402

Pay as you go,
no subscription,
no KYC.

for people and agents accessing AI models —
one endpoint, priced per request,
settled onchain with no geo-restriction.

Read the API reference

Fable 5

router/claude-fable-5

x402

CONTEXT 1M

IN $10.00 / 1M · OUT $50.00 / 1M

SETTLEMENT [USDG/ROBINHOOD]

1 of 400+ in the catalog

GPT-5.6 Sol

router/gpt-5.6-sol

x402

CONTEXT 400K

IN $5.00 / 1M · OUT $30.00 / 1M

SETTLEMENT [USDG/ROBINHOOD]

1 of 400+ in the catalog
402 PAYMENT REQUIRED

Your request hits the router without a key. The router answers 402, quotes the price in USDG, your agent signs a gasless EIP-3009 authorization and re-sends, a facilitator settles it onchain, and the response streams back. No account, no invoice, no float.

A single doorevery model

Unified API

POSTapi.…/v1/inference

OpenAI-compatible. Point your existing SDK at one base URL and every model in the catalog is reachable by name. When a provider deprecates a model, the router re-points the alias — your code does not move.

models
400+
providers
3
settle chains
19

Vendors change
the endpoint doesn't

LLM providers run at 99–99.5% uptime — 6 to 14× worse than cloud infrastructure. Two providers with automatic failover cut expected downtime from about 41 hours a year to about 12 minutes.

TianPan.co, LLM API Resilience in Production · 2026-03-11

001

Dolphin

free

Open-weight models served at no charge. Rate-limited, never metered.

billingFree

002

Surplus

x402

Frontier models at list price. Each call settles in USDG before the stream opens.

billingPay per request

003

AntSeed

x402

Redundant capacity for the same catalog — the fallback when an upstream degrades.

billingPay per request
Many routes converging into a single bright node and continuing as one beam

Routing

3 upstreams, 400+ models, one endpoint.

A request names a model, not a vendor. Where it lands is a routing decision the router makes on price, latency and whether the upstream is currently answering.

One catalog
priced per request

Text, image, video and audio models from every provider, merged into one list with one price column. Free tiers stay free — you pay only where the upstream charges.

ModelProviderContextPrice
Claude Fable 5
$10.00 in · $50.00 out / 1M
Surplus1000K$50.00
Llama 4.2 Maverick
no charge
Dolphin1000KFree
FLUX 2 Pro
image · image
Surplus4K$0.05 image
Kling 3.0
video · second of video
Surplus2K$0.085 second of video
GPT-5.6 Luna
$0.20 in · $1.20 out / 1M
Surplus400K$1.20

5 of 400+ models in the merged catalog

GET /v1/inference/models
pay as you gopay as you gopay as you gopay as you go
no subscriptionno KYCno sign-upno geo-restrictionsno censorshipany model, any countryno subscriptionno KYCno sign-upno geo-restrictionsno censorshipany model, any country

Every modality
one settlement rail

001 · Text

Fable 5

router/claude-fable-5

x402
Summarize this contract
Three clauses need review — here's the diff.
Draft the reply
1M contextIn $10.00 / 1MOut $50.00 / 1M

002 · Image

Z-Image Turbo

router/z-image-turbo

x402
A generated frame from the image model, shown as it is returned by the routerdurable media URL
Agent-persistableDurable media URL~$0.005 / image

003 · Video

Kling 3.0

router/kling-v3

x402
A still from a five-second clip generated through the router
0:00
0:05
5s cinematicSeconds billed$0.07–$0.10 / sec

004 · Audio

Full-track audio

router/audio-v4

x402
track_01.wav0:42 / 3:18
Full-trackStem exportAgent-ready

Own your conversations.
Own your AI.

Threads live in your browser, not a company vault. Search past sessions, start new ones, and pick up where you left off without anyone else holding the transcript.

local-first historyno server transcriptexport anytime
inference.…
Ask what they won't let through.
No corporate filter. Your prompt routes across independent nodes, not one company's moderation stack.
Ask anything…↗

The rail is real,
and it is dated.

Every number below carries its source. Nothing here is rounded up, and the one claim that needs a qualifier keeps it.

The x402 Foundation went operationally live under the Linux Foundation with 40 member organizations.
Members include Visa, Mastercard, American Express, Stripe, Google, AWS, Shopify, Cloudflare and Coinbase.

Linux Foundation / PRNewswire · 2026-07-14

Coinbase reported roughly 165 million x402 transactions and about $50M in cumulative volume.
Measured to late April 2026, across the networks the hosted facilitator supports.

Coinbase figures via tftc.io · 2026-04-30

Robinhood Chain went live on mainnet on 1 July 2026 and topped the RWA.xyz Networks leaderboard by 27 July.
USDG, its first natively issued stablecoin, is the chain's primary settlement asset.

Arbitrum blog; RWA.xyz via CryptoTimes · 2026-07-27

Inference-as-a-service is a $23.4 billion market in 2026, forecast to reach $197.5 billion by 2035.
The broader AI inference market — including on-prem hardware and edge silicon — reaches $117.8B in 2026.

Precedence Research; Fortune Business Insights · 2026-07-13